Gartner analyst Merv Adrian recently highlighted some of the recent movements in Hadoop Land, with several companies introducing products "intended to improve Hadoop speed."
This seems odd, as that wouldn't be my top pick for how to improve Hadoop or, really, most of the Big Data technologies out there. By many accounts, the biggest need in Hadoop is improved ease of use, not improved performance, something Adrian himself confirms:
@mjasay true that. E.g., what good is SQL if there's no metadata there yet?
— Merv Adrian (@merv) May 1, 2013
Hadoop already delivers exceptional performance on commodity hardware, compared to its stodgy proprietary competition. Where it's still lacking is in ease of use.
Not that Hadoop is alone in this.
Today, despite the information deluge, enterprise decision makers are often unable to access the data in a useful way. The tools are designed for those who speak the language of algorithms and statistical analysis. It’s simply too hard for the everyday user to “ask” the data any questions – from the routine to the insightful. The end result? The speed of big data moves at a slower pace … and the power is locked in the hands of the few.
Lucas goes on to argue that the solution to the data scientist shortage is to take the science out of data science; that is, consumerize Big Data technology such that non-PhD-wielding business people can query their data and get back meaningful results.
The Value Of Open Source To Deciphering Big Data
Perhaps. But there's actually an intermediate step before we reach the Promised Land of full consumerization of Big Data. It's called open source.
Even with technology like Hadoop that is open source yet still too complex, the benefits of using Hadoop far outweigh the costs (financial and productivity-wise) associated with licensing an expensive data warehousing or analytics platform. As Alex Popescu writes, Hadoop "allows experimenting and trying out new ideas, while continuing to accumulate and storing your data. It removes the pressure from the developers. That’s agility."
But these benefits aren't unique to Hadoop. They're inherent in any open-source project. Now imagine we could get open-source software that fits our Big Data needs and is exceptionally easy to use plus is almost certainly already being used within our enterprises...? That is the promise of MongoDB, consistently cited as one of the industry's top-two Big Data technologies. MongoDB makes it easy to get started with a Big Data project.
Using MongoDB To Innovate
Consider the City of Chicago. The Economist wrote recently about the City of Chicago's predictive analytics platform, WindyGrid. What The Economist didn't mention is that WindyGrid started as a pet project on chief data officer Brett Goldstein's laptop. Goldstein started with a single MongoDB node, and iterated from there, turning it into one of the most exciting data-driven applications in the industry today.
Given that we often don't know exactly which data to query, or how to query, or how to put data to work in our applications, this is precisely how a Big Data project should work. Start small, then iterate toward something big. This kind of tinkering simply is difficult to impossible with a relational database, as The Economist's Kenneth Cukier points out in his book, Big Data: A Revolution That Will Transform How We Live, Work, and Think:
Conventional, so-called relational, databases are designed for a world in which data is sparse, and thus can be and will be curated carefully. It is a world in which the questions one wants to answer using the data have to be clear at the outset, so that the database is designed to answer them - and only them - efficiently.
But with a flexible document database like MongoDB, it suddenly becomes much easier to iterate toward Big Data insights. We don't need to go out and hire data scientists. Rather, we simply need to apply existing, open-source technology like MongoDB to our Big Data problems, which jibes perfectly with Gartner analyst Svetlana Sicular's mantra that it's easier to train existing employees on Big Data technologies than it is to train data scientists on one's business.
Except, in the case of MongoDB, odds are that enterprises are already filled with people that understand MongoDB, as 451 Research's LinkedIn analysis suggests:
In sum, Big Data needn't be daunting or difficult. It's a download away.